Gaussian distribution maximizes entropy for given covariance
Theorem 1
Let be a continuous probabilitydensity function. Let be random variables
with density and with covariance matrix , . Let be the distribution
of themultidimensional Gaussian (http://planetmath.org/JointNormalDistribution) with mean and covariance matrix . Then the Gaussian distribution maximizes the differential entropy for a given covariance matrix . That is, .